A Data-Driven Approach to Cancer Prognosis Using KNN Algorithm

Abstract

A vital component of medical research is cancer prognosis, which facilitates early diagnosis and efficient treatment planning. This study offers a data-driven method for accurately predicting the prognosis of cancer using the K-Nearest Neighbors (KNN) algorithm. With an overall accuracy of 94% after being trained and assessed on a structured dataset, the model proved to be dependable in categorization. Surgery had the best precision (0.97) of any class, showing a high capacity to accurately identify patients with few false positives. The model's efficacy was further confirmed by the classification report and confusion matrix, which displayed strong recall and F1-scores across many categories. According to these results, KNN may be a useful tool for helping medical practitioners make well-informed decisions about the prognosis of cancer. This study presents a novel application of the KNN algorithm for multi-modal cancer treatment prognosis, demonstrating high predictive accuracy and offering data-driven support for clinical decision-making. To further improve prediction performance, future research may concentrate on feature selection, hyperparameter tuning, and hybrid models. This study demonstrates the promise of machine learning in medical diagnostics by offering a successful and non-invasive method for predicting the prognosis of cancer.